|Publication number||US6999651 B2|
|Application number||US 10/442,515|
|Publication date||Feb 14, 2006|
|Filing date||Apr 4, 2003|
|Priority date||Mar 30, 2001|
|Also published as||US6678437, US20020141689, US20050058392|
|Publication number||10442515, 442515, US 6999651 B2, US 6999651B2, US-B2-6999651, US6999651 B2, US6999651B2|
|Inventors||Dahong Qian, Marc Hertzberg, Wayne Wong, Amit Burstein|
|Original Assignee||Analog Devices, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (6), Referenced by (4), Classifications (17), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application is a division of U.S. patent application Ser. No. 09/823,811, filed Mar. 30, 2001, now U.S. Pat. No. 6,678,437.
The invention is concerned with the field of fiber optical communication architectures and network systems, and the problems of optically switching fiber-carried light signals between input and output fibers in high speed optical communication networks and the like, with more particular concern for optical cross-connect switching and control between fiber optical input and output (I/O) bundle arrays.
The switching of signal light beams between optical fiber bundles carrying data optical I/O cards in communication networks and systems is currently being addressed with the aid of optical microelectric-electromechanical devices (MEMS or MOEMS) carrying tiltable micro-mirrors orientable by electrical signal control to deflect incident light beams along desired paths—switching the output light from input fibers via reflection from the micro-mirrors to predetermined fibers in an output fiber bundle—the so-called optical cross-connect switching mentioned above.
Examples of such MEMS devices are disclosed, for example, in U.S. Pat. Nos. 6,052,498; 6,147,876; 6,150,724; and 6,151,173; and details of such optical switching systems are described, for example, in the following references:
In general, such free-space optical switches steer the input light from any one of input fibers to any one of the output fibers. The number of input fibers and output fibers can range from tens to thousands. The principal challenges to controlling the light are providing for:
In the implementation of, for example, the optical switches of the above-cited Slater et al reference  and other similar prior art systems, electrical signal feedback circuits for establishing precise mirror position are widely used.
A piezoresistive sensing mechanism is generally provided on the torsional mirror support to measure position of the mirror. This sensing signal is fed to an error amplifier with a command input from digital signal processor a (DSP)-controlled digital analog controller (DAC). The error signal is fed through the analog controller to control the actuation or control voltage signal orienting the mirror. Problems and limitations with this electrical feedback approach, however, arise from the facts that the controller is implemented in analog components which are not flexible; the controller is not scalable because each mirror needs its control; and the DSP is also not scalable because it is tied to the controller all of the time.
Resort has therefore been taken, as described in the Bishop reference  above, to the use of external optical feedback without using such on-chip electrical feedback signals—i.e. using only port card optical power feedback. Since such involve a very slow loop, however, this poses extremely high requirements on MEMS design, in that the resonant frequency has to be much higher than the disturbance. The optical power feedback from the I/O cards, furthermore, cannot give directional information. This thus requires extra time and computation power to adjust the mirror beam positions. While additional optical sensing schemes can be built into the optical box as in the Bishop proposal, this further increases the complexity of the optical box and adds more variables to control, i.e. the relative positions and alignment of the optical sensing elements in the optical box.
The present invention, on the other hand, through a novel use of intermediate node monitoring and with the use of test and real traffic data path switching, has admirably overcome the above-described and other limitations of such prior art techniques, as will later be more fully explained.
An object of the present invention, accordingly, is to provide a new and improved approach, method, architecture and system for optically switching light communication signals in fiber optic networks and the like that obviate the above and other prior art limitations and problems, and, indeed, provide for faster switching and optical path connections than have heretofore been available with optical cross-connect and other switching techniques.
A further object is to provide such improvement with novel working and test protection switching at intermediate nodes of the network that reduce overall connection setup time.
Other and further objects are stressed in the descriptions that follow and are more particularly pointed out in the appended claims.
In summary, however, from one of its broader viewpoints, the invention embraces in an optical path switching system wherein light beam data traffic is to be switched by MEMS orientable mirrors between a source node and a destination node through intermediate nodes having ingress and egress ports that are to be connected to set up the desired optical path connection between the source and destination nodes and under the control of a protocol control plane, a method of reducing the overall switch connection setup time while guaranteeing the integrity of the connection before the actual propagating of the data traffic from the source, that comprises, transmitting a test light beam from the ingress port of each intermediate node to a test detector at its egress port upon the identifying from the control plane of the ingress and egress port numbers of the node, thereby to set up and verify the integrity of an optical path connection therebetween; thereupon terminating the test light beam transmission; and then awaiting the subsequent propagation of the data traffic from the source along the setup-connected optical path to the selected destination node.
Preferred and best mode designs and embodiments are hereinafter explained in detail.
The invention will now be described in connection with the accompanying drawings,
FIGS. 8(A)–(E) show successive operational steps in the micro-mirror “shaker” algorithmic flow;
As before mentioned, it has been previously proposed to employ optical cross-connect switching configurations, as of the type shown in
In such and other prior art systems, moreover, switching connections have to await final micro-mirror adjustments and then the arrival of the actual or real or working data traffic before starting the switching connections.
Through the use of the novel connection setup and provisioning signaling flow of
Considering, now, the details of this connection setup flow, as shown in
Once such connection is set up and verified, it will be reported to the upper level control plane. Then, the 2×2 switches are switched back to the normal position shown in
A more detailed explanation of such setup connections is provided by the flow chart of
Each mirror in the optical switch is controlled by multiple control loops, with each loop accomplishing different goals. By combining the control loops, the requirements of speed and accuracy are met. The inner electrical position feedback loop Fe,
To meet the control challenges outlined above, the beam moving and holding control systems are preferably partitioned to increase scalability. The “Mover” algorithm for coarsely moving the mirror to steer a beam from one fiber to a different output fiber is, however, relatively complex and difficult. As for the “Shaker” algorithm, the job of keeping the beam pointed at a particular output fiber is considerably simpler.
Further details of such “Shaker” FPGA algorithm flow are presented in FIGS. 8(A)–(E).
The sensor data is collected into sensor data memory and is the sensor data is sent to the “Mover” algorithm. As earlier mentioned, the “Shaker” algorithm runs on all sensor data (even disabled servos and servos under “Mover” control). If the position command to the “Shaker” algorithm is indicating a disabled servo or “Mover” control, this causes all history memories and also actuation memory to be zeroed when the “Shaker” algorithm executes. Thus, there is no need to specifically initialize state when it is desired to enable the “Shaker” algorithm. The “Mover” algorithm actuation data arriving are written into the actuation memory,
An alternative digital control system (DCS) for the MEMS micro-mirror arrays is presented in
The basic operation is to sample a position sensor input for the channel, execute the control algorithm to generate an output based on that input and previous history, and apply the output to the channel. This basic operation is to be executed many times in parallel in many sets of inputs and outputs in a time slice fashion. The number of channels (twice the number of mirrors to be controlled), the required sample rate, and the amount of processing time needed by each iteration of the control algorithm determines the number of DSP processors required. The design allows for easily changing the number of DSPs being used.
A single A/D converter (“A/D” in
The single D/A converter shared for all output channels provides a S/H (sample and hold) for each channel between the D/A converter and the driver for the channel. Since the S/Hs may be formed by a simple circuit such as an analog switch followed by a hold capacitor, they may have a droop rate which requires them to be refreshed periodically. The refresh rate requirement of the S/H circuits may be of higher frequency than the frequency at which each channel is visited by the control algorithm, and the design provides for this possibility.
In order to process the maximum number of channels per DSP, external circuitry is responsible for coordinating the selection of the input channel, the command to initiate the A/D conversion, and the placing of the conversion results into the appropriate memory location in a correct DSP. The DSPs are only responsible for running the controller algorithms and writing the output to the timing controller. The DSPs are designed to initiate the output to the timing controller so that the watchdogs can observe proper behavior. It should be noted that the DSPs also do not need to communicate directly with the outside world over the command/status interface. The “Auxiliary uP”,
Rather than having the DSPs directly writing to the D/A and selecting the S/Hs, moreover, a timing and refresh controller is inserted between the DSPs and the D/A (top of
If a DSP fails and stops writing new control output values to the memory, furthermore, the last values written are maintained and applied to the S/Hs for the affected channels by the hardware refresh operation. This keeps the mirrors from losing their commands while the affected DSP is re-initialized.
Because the responsibilities of the DSPs have been stripped down to an absolute minimum, their operation is very predictable. They receive a periodic signal from the timing controller which causes them to run a fixed-length controller calculation. After the calculation, they output the result of the calculation by writing it to the timing controller. The watchdogs are thus able to watch that the calculated value emerges from the DSPs at exactly the expected time slot. If a problem is detected by the watchdogs, they inform the before-mentioned Auxiliary uP, which can then attempt to recover from the fault by re-starting the affected DSP. While this is happening, the mirror position is held by the timing controller hardware refresh feature.
The Auxiliary uP also offloads the DSPs from any responsibilities other than in servo calculations. This processor accepts commands and provides status to the outside world over the communications interface. To the extent that information from the outside world needs to be communicated to and from the DSPs, the uP transfers this information by way of DMA into the internal memories of the DSP. Information which may need to be transferred includes desired deflection angle for a particular channel (transferred to a DSP) and servo performance status and statistics (transferred from a DSP).
The “FLASH” memory of
The details of the deflection memory implementation of the system are shown in
This accomplishes both primer position forcing and maintaining (refreshing) functions by simple and homogeneous hardware.
Advantages of this alternate DCS approach reside in the fact that the DSP devices perform only servo calculations and are relieved of almost all tasks not associated with actual calculations. Means is provided for detecting and recovering from a DSP fault, and the refreshing of control signals to the mirrors is independent of DSP operation.
In accordance with the present invention, accordingly, a digital controller system is provided for the MEMS micro-mirrors based on a combination of two control loops—an electrical sensing loop (using capacitive, coarse control), and an optical power feedback loop (fine control). A separate control is provided for moving a mirror to a designated position from the control for holding and maintaining the mirror at its precise position. This increases the scalability of the control hardware and software. After connections are set up successfully, moreover, beam position optimization is continually performed in the background without interruption of the connection. This optimization is done by small incremental adjustments of the mirror positions while monitoring optical power fluctuation. This optimization is crucial to guarantee that the next mirror moving will be captured by another fiber; i.e. the calibration table is always optimized. The invention, moreover, uses test ports to set up paths even before the real data traffic answers, in order to reduce path set up time and guarantee the connectivity. This is also used regularly to exercise the mirrors which are not used by the current connections to compensate for any shift.
Further modifications will also occur to those skilled in this art and such are considered to fall within the scope and spirit of the invention as defined in the appended claims.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US6360035 *||Jun 2, 2000||Mar 19, 2002||Seagate Technology Llc||Optical head using micro-machined elements|
|US6483962 *||May 24, 2001||Nov 19, 2002||Vlad J. Novotny||Optical cross connect switching array system with optical feedback|
|US6587611 *||Jun 6, 2000||Jul 1, 2003||Calient Networks, Inc.||Maintaining path integrity in an optical switch|
|US6625341 *||Jun 12, 2001||Sep 23, 2003||Vlad J. Novotny||Optical cross connect switching array system with electrical and optical position sensitive detection|
|US6728016 *||Jun 5, 2000||Apr 27, 2004||Calient Networks, Inc.||Safe procedure for moving mirrors in an optical cross-connect switch|
|US20020092963 *||Oct 19, 2001||Jul 18, 2002||Domash Lawrence H.||Semitransparent sensor for steering an optical beam|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US9091820||Jun 10, 2013||Jul 28, 2015||Freescale Semiconductor, Inc.||Communication system die stack|
|US9094135||Jun 10, 2013||Jul 28, 2015||Freescale Semiconductor, Inc.||Die stack with optical TSVs|
|US9261556||Jun 10, 2013||Feb 16, 2016||Freescale Semiconductor, Inc.||Optical wafer and die probe testing|
|US20060215244 *||May 31, 2006||Sep 28, 2006||Jacob Yosha||Vehicle display system|
|U.S. Classification||385/18, 385/15, 385/17|
|International Classification||H04Q11/00, G02B6/26, G02B6/35|
|Cooperative Classification||H04Q11/0005, G02B6/3588, H04Q2011/003, H04Q2011/0043, H04Q2011/0083, H04Q2011/0039, G02B6/3512, G02B6/3556, G02B6/359|
|European Classification||H04Q11/00P2, G02B6/35R2|
|Aug 14, 2009||FPAY||Fee payment|
Year of fee payment: 4
|Mar 13, 2013||FPAY||Fee payment|
Year of fee payment: 8
|Apr 19, 2013||SULP||Surcharge for late payment|